forked from commaai/speedchallenge
-
Notifications
You must be signed in to change notification settings - Fork 1
/
loader.py
147 lines (122 loc) · 5.68 KB
/
loader.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
import tensorflow as tf
import numpy as np
image_feature_description = {
'frame_one': tf.io.FixedLenFeature([], tf.string),
'frame_two': tf.io.FixedLenFeature([], tf.string),
'plus_one_position': tf.io.FixedLenFeature([3], tf.float32),
'plus_one_orientation': tf.io.FixedLenFeature([3], tf.float32),
# 'lead_car_one': tf.io.FixedLenFeature([3], tf.float32),
# 'lead_car_two': tf.io.FixedLenFeature([3], tf.float32),
'pitch': tf.io.FixedLenFeature([], tf.float32),
'yaw': tf.io.FixedLenFeature([], tf.float32),
'speed': tf.io.FixedLenFeature([], tf.float32),
}
def normalize_image(image):
"""Normalize the image to zero mean and unit variance."""
# The image normalization is identical to Cloud TPU ResNet.
mean_rgb = [0.485 * 255, 0.456 * 255, 0.406 * 255]
stddev_rgb = [0.229 * 255, 0.224 * 255, 0.225 * 255]
image = tf.cast(image, dtype=tf.float32)
image -= tf.constant(mean_rgb, shape=(1, 1, 3), dtype=tf.float32)
image /= tf.constant(stddev_rgb, shape=(1, 1, 3), dtype=tf.float32)
return image
def decode_and_process_frame(frame, mirror, training):
image = tf.image.decode_and_crop_jpeg(frame, [112, 0 , 256, 640], channels=3, fancy_upscaling=False, dct_method="INTEGER_FAST")
# image = tf.image.decode_and_crop_jpeg(frame, [142, 0, 197, 640], channels=3, fancy_upscaling=False, dct_method="INTEGER_FAST")
# image = tf.image.resize(image, [128, 416])
if training:
# image = tf.image.random_hue(image, 0.08)
# image = tf.image.random_saturation(image, 0.6, 1.6)
# image = tf.image.random_brightness(image, 0.05)
# image = tf.image.random_contrast(image, 0.7, 1.3)
# image = cutout(image, 40, 0)
pass
if mirror:
image = tf.image.flip_left_right(image)
# image = tf.image.convert_image_dtype(image, tf.float32)
# image = tf.image.per_image_standardization(image)
image = normalize_image(image)
return image
def cutout(image, pad_size, replace=0):
"""Apply cutout (https://arxiv.org/abs/1708.04552) to image.
This operation applies a (2*pad_size x 2*pad_size) mask of zeros to
a random location within `img`. The pixel values filled in will be of the
value `replace`. The located where the mask will be applied is randomly
chosen uniformly over the whole image.
Args:
image: An image Tensor of type uint8.
pad_size: Specifies how big the zero mask that will be generated is that
is applied to the image. The mask will be of size
(2*pad_size x 2*pad_size).
replace: What pixel value to fill in the image in the area that has
the cutout mask applied to it.
Returns:
An image Tensor that is of type uint8.
"""
image_height = tf.shape(image)[0]
image_width = tf.shape(image)[1]
# Sample the center location in the image where the zero mask will be applied.
cutout_center_height = tf.random.uniform(
shape=[], minval=0, maxval=image_height,
dtype=tf.int32)
cutout_center_width = tf.random.uniform(
shape=[], minval=0, maxval=image_width,
dtype=tf.int32)
lower_pad = tf.maximum(0, cutout_center_height - pad_size)
upper_pad = tf.maximum(0, image_height - cutout_center_height - pad_size)
left_pad = tf.maximum(0, cutout_center_width - pad_size)
right_pad = tf.maximum(0, image_width - cutout_center_width - pad_size)
cutout_shape = [image_height - (lower_pad + upper_pad),
image_width - (left_pad + right_pad)]
padding_dims = [[lower_pad, upper_pad], [left_pad, right_pad]]
mask = tf.pad(
tf.zeros(cutout_shape, dtype=image.dtype),
padding_dims, constant_values=1)
mask = tf.expand_dims(mask, -1)
mask = tf.tile(mask, [1, 1, 3])
image = tf.where(
tf.equal(mask, 0),
tf.ones_like(image, dtype=image.dtype) * replace,
image)
return image
@tf.function
def parse_record(tfrecord, training):
proto = tf.io.parse_single_example(tfrecord, image_feature_description)
if training:
mirror = tf.random.uniform([]) < 0.5
else:
mirror = False
frame_one = decode_and_process_frame(proto['frame_one'], mirror, training)
frame_two = decode_and_process_frame(proto['frame_two'], mirror, training)
position = proto['plus_one_position']
orienation = proto['plus_one_orientation']
speed = proto['speed']
# lead_car_one = proto['lead_car_one']
# lead_car_two = proto['lead_car_two']
if mirror:
position = (1, -1, 1) * position
orienation = (-1, 1, -1) * orienation
# lead_car_one = (1, -1, 1) * lead_car_one
# lead_car_two = (1, -1, 1) * lead_car_two
pose = tf.concat((position, orienation), axis=0)
pitch = proto['pitch']
yaw = proto['yaw']
# if not training or tf.random.uniform([]) < 0.5:
image = tf.concat((frame_one, frame_two), axis=2)
return {'frames': image}, {'pose': pose, 'speed': [speed], 'pitch': [pitch]}
# else:
# rev_image = tf.concat((frame_two, frame_one), axis=2)
# rev_pose = -1 * pose
# return {'frames': rev_image}, {'pose': rev_pose, 'speed': [speed]}
def load_tfrecord(file_pattern, batch_size, training):
dataset = tf.data.Dataset.list_files(file_pattern, shuffle=training)
def _prefetch_data(filename):
dataset = tf.data.TFRecordDataset(filename).prefetch(1)
return dataset
dataset = dataset.interleave(_prefetch_data, cycle_length=100, num_parallel_calls=tf.data.experimental.AUTOTUNE)
if training:
dataset = dataset.shuffle(batch_size * 128)
dataset = dataset.map(lambda x: parse_record(x, training), num_parallel_calls=tf.data.experimental.AUTOTUNE)
dataset = dataset.batch(batch_size)
dataset = dataset.prefetch(buffer_size=tf.data.experimental.AUTOTUNE)
return dataset